Causal Structure Learning by Using Intersection of Markov Blankets

TL;DR

This paper presents EEMBI, a new causal structure learning algorithm that leverages the intersection of Markov blankets, and its extension EEMBI-PC, which incorporates the PC algorithm for improved causal discovery.

Contribution

The paper introduces EEMBI and EEMBI-PC, novel algorithms that combine Bayesian networks, SCMs, and the PC algorithm for enhanced causal structure learning.

Findings

EEMBI effectively identifies causal structures using Markov blanket intersections.

EEMBI-PC improves causal discovery accuracy by integrating PC algorithm steps.

The methods bridge Bayesian networks and SCMs for robust causal inference.

Abstract

In this paper, we introduce a novel causal structure learning algorithm called Endogenous and Exogenous Markov Blankets Intersection (EEMBI), which combines the properties of Bayesian networks and Structural Causal Models (SCM). Furthermore, we propose an extended version of EEMBI, namely EEMBI-PC, which integrates the last step of the PC algorithm into EEMBI.